Overview
Resizes an image to any dimensions. Use it for non-integer scaling such as
48ร48 โ 32ร32, which integer-only pixel_scale cannot handle.
Usage tips
- Choose among the three resize tools by intent: integer scaling =
pixel_scale(preserves pixels), fitting dimensions by adding or removing margins =canvas_resize/fit_to_canvas, stretching the artwork itself to arbitrary dimensions = Content Resize. - Non-integer downscaling inevitably distorts pixel art. Treat this as a tool for cases where matching the dimensions matters more than preserving the pixels, such as importing external assets.
- Use
NearestNeighbor(the default) for pixel art.Bilinearis for preparing photographic material. It interpolates translucent edges in premultiplied-alpha space, preventing hidden RGB in transparent pixels from creating dark or colored fringes.
Related nodes
pixel_scaleโ lossless integer scalingcanvas_resize/fit_to_canvasโ match dimensions through canvas operationsimage_loadโ import external assets (the main upstream node for this one)
Sampling modes
| Method | Label | Description |
|---|---|---|
| NearestNeighbor | Nearest neighbor | Crisp edges; ideal for pixel art |
| Bilinear | Bilinear | Smooth interpolation; suited to photos and natural images |
Technical details
- Same dimensions: passes the image through without copying it, which is fast
- Nearest neighbor: takes the nearest pixel as-is; works with non-integer scaling
- Bilinear: interpolates the four surrounding pixels in premultiplied-alpha space, producing intermediate colors without exposing hidden RGB; fully transparent results are canonical transparent black
Examples
Change the resolution from 32 to 64px
Scale 32ร32 pixel art up with width: 64, height: 64, and sampling: NearestNeighbor.
Non-integer resizing
Scale a 48ร48 image down to width: 32, height: 32 (a non-integer ratio that Pixel Scale cannot handle).
Differences from PixelScale
| Item | ContentResize | PixelScale |
|---|---|---|
| Size specification | Arbitrary (width ร height) |
Integer factors only (รN / รทN) |
| Sampling | NN or Bilinear | NN only |
| Purpose | Convert to arbitrary dimensions | Dedicated to integer upscaling and downscaling |